Add icu-locale macros

Cargo.toml

@@ -8,6 +8,7 @@ members = [
     "components/icu4x",
     "components/uniset",
     "components/locale",
+    "components/locale/macros",
     "components/num-util",
     "components/pluralrules",
 ]

components/locale/macros/Cargo.toml

@@ -0,0 +1,21 @@
+[package]
+name = "icu-locale-macros"
+description = "proc-macros for icu-locale"
+version = "0.0.1"
+authors = ["The ICU4X Project Developers"]
+edition = "2018"
+readme = "README.md"
+repository = "https://github.com/unicode-org/icu4x"
+license-file = "../../LICENSE"
+categories = ["internationalization"]
+include = [
+    "src/**/*",
+    "Cargo.toml",
+    "README.md"
+]
+
+[lib]
+proc_macro = true
+
+[dependencies]
+icu-locale = { version = "0.0.1", path = "../" }

components/locale/macros/src/lib.rs

@@ -0,0 +1,109 @@
+extern crate proc_macro;
+
+use proc_macro::TokenStream;
+
+fn get_value_from_token_stream(input: TokenStream) -> String {
+    let val = format!("{}", input);
+    if !val.starts_with('"') || !val.ends_with('"') {
+        panic!("Argument must be a string literal.");
+    }
+    let len = val.len();
+    (&val[1..len - 1]).to_string()
+}
+
+fn language_output(input: Option<u64>) -> String {
+    let val = if let Some(raw) = input {
+        format!("Some({})", raw)
+    } else {
+        "None".to_string()
+    };
+    format!("unsafe {{ icu_locale::subtags::Language::from_raw_unchecked({}) }}", val)
+}
+
+fn script_output(input: u32) -> String {
+    format!("unsafe {{ icu_locale::subtags::Script::from_raw_unchecked({}) }}", input)
+}
+
+fn region_output(input: u32) -> String {
+    format!("unsafe {{ icu_locale::subtags::Region::from_raw_unchecked({}) }}", input)
+}
+
+fn variant_output(input: u64) -> String {
+    format!("unsafe {{ icu_locale::subtags::Variant::from_raw_unchecked({}) }}", input)
+}
+
+#[proc_macro]
+pub fn language(input: TokenStream) -> TokenStream {
+    let val = get_value_from_token_stream(input);
+
+    let parsed: icu_locale::subtags::Language = val.parse().expect("Malformed Language Subtag");
+
+    let lang = language_output(parsed.into_raw());
+
+    lang.parse().unwrap()
+}
+
+#[proc_macro]
+pub fn script(input: TokenStream) -> TokenStream {
+    let val = get_value_from_token_stream(input);
+
+    let parsed: icu_locale::subtags::Script = val.parse().expect("Malformed Script Subtag");
+
+    let script = script_output(parsed.into_raw());
+
+    script.parse().unwrap()
+}
+
+#[proc_macro]
+pub fn region(input: TokenStream) -> TokenStream {
+    let val = get_value_from_token_stream(input);
+
+    let parsed: icu_locale::subtags::Region = val.parse().expect("Malformed Region Subtag");
+
+    let region = region_output(parsed.into_raw());
+
+    region.parse().unwrap()
+}
+
+#[proc_macro]
+pub fn variant(input: TokenStream) -> TokenStream {
+    let val = get_value_from_token_stream(input);
+
+    let parsed: icu_locale::subtags::Variant = val.parse().expect("Malformed Variant Subtag");
+
+    let variant = variant_output(parsed.into_raw());
+
+    variant.parse().unwrap()
+}
+
+#[proc_macro]
+pub fn langid(input: TokenStream) -> TokenStream {
+    let val = get_value_from_token_stream(input);
+
+    let langid: icu_locale::LanguageIdentifier = val.parse().expect("Malformed Language Identifier");
+
+    let lang = language_output(langid.language.into_raw());
+
+    let script = if let Some(script) = langid.script {
+        format!("Some({})", script_output(script.into_raw()))
+    } else {
+        "None".to_string()
+    };
+    let region = if let Some(region) = langid.region {
+        format!("Some({})", region_output(region.into_raw()))
+    } else {
+        "None".to_string()
+    };
+    let variants = format!("icu_locale::subtags::Variants::from_vec_unchecked(vec![])");
+
+    let output = format!(r#"
+icu_locale::LanguageIdentifier {{
+    language: {},
+    script: {},
+    region: {},
+    variants: {},
+}}
+"#, lang, script, region, variants);
+
+    output.parse().unwrap()
+}

components/locale/macros/tests/macros.rs

@@ -0,0 +1,54 @@
+use icu_locale_macros::*;
+use icu_locale::subtags;
+use icu_locale::LanguageIdentifier;
+
+const LANG_PL: subtags::Language = language!("pL");
+const SCRIPT_LATN: subtags::Script = script!("lAtN");
+const REGION_US: subtags::Region = region!("us");
+const VARIANT_MACOS: subtags::Variant = variant!("MACOS");
+const LANGID: LanguageIdentifier = langid!("de-Arab-AT");
+
+#[test]
+fn language() {
+    let lang = language!("Pl");
+
+    assert_eq!(lang, "pl");
+    assert_eq!(LANG_PL, "pl");
+    assert_eq!(lang, LANG_PL);
+}
+
+#[test]
+fn script() {
+    let script = script!("latn");
+
+    assert_eq!(script, "Latn");
+    assert_eq!(SCRIPT_LATN, "Latn");
+    assert_eq!(script, SCRIPT_LATN);
+}
+
+#[test]
+fn region() {
+    let region = region!("us");
+
+    assert_eq!(region, "US");
+    assert_eq!(REGION_US, "US");
+    assert_eq!(region, REGION_US);
+}
+
+#[test]
+fn variant() {
+    let variant = variant!("macOS");
+
+    assert_eq!(variant, "macos");
+    assert_eq!(VARIANT_MACOS, "macos");
+    assert_eq!(variant, VARIANT_MACOS);
+}
+
+#[test]
+fn langid() {
+    let langid = langid!("de_Arab_aT");
+
+    assert_eq!(langid.to_string(), "de-Arab-AT");
+    assert_eq!(LANGID.to_string(), "de-Arab-AT");
+    assert_eq!(langid, LANGID);
+}

components/locale/src/parser/langid.rs

@@ -87,7 +87,7 @@ pub fn parse_language_identifier_from_iter<'a>(
         language,
         script,
         region,
-        variants: subtags::Variants::from_vec_unchecked(variants),
+        variants: subtags::Variants::from_raw_unchecked(Some(variants.into_boxed_slice())),
     })
 }
 

components/locale/src/subtags/language.rs

@@ -67,6 +67,53 @@ impl Language {
         }
     }
 
+    /// Deconstructs the `Language` into raw format to be consumed
+    /// by `from_raw_unchecked`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use icu_locale::subtags::Language;
+    ///
+    /// let lang = Language::from_bytes(b"en")
+    ///     .expect("Parsing failed.");
+    ///
+    /// let raw = lang.into_raw();
+    /// let lang = unsafe { Language::from_raw_unchecked(raw) };
+    /// assert_eq!(lang, "en");
+    /// ```
+    pub fn into_raw(self) -> Option<u64> {
+        self.0.map(|v| v.into())
+    }
+
+    /// Constructor which takes a raw value returned by
+    /// `into_raw`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use icu_locale::subtags::Language;
+    ///
+    /// let lang = Language::from_bytes(b"en")
+    ///     .expect("Parsing failed.");
+    ///
+    /// let raw = lang.into_raw();
+    /// let lang = unsafe { Language::from_raw_unchecked(raw) };
+    /// assert_eq!(lang, "en");
+    /// ```
+    ///
+    /// # Safety
+    ///
+    /// This function accepts any u64 that is exected to be a valid
+    /// `TinyStr8` and a valid `Language` subtag.
+    pub const unsafe fn from_raw_unchecked(v: Option<u64>) -> Self {
+        if let Some(v) = v {
+            Self(Some(TinyStr8::new_unchecked(v)))
+        } else {
+            Self(None)
+        }
+    }
+
     /// A helper function for displaying
     /// a `Language` subtag as a `&str`.
     ///

components/locale/src/subtags/region.rs

@@ -57,6 +57,48 @@ impl Region {
         }
     }
 
+    /// Deconstructs the `Region` into raw format to be consumed
+    /// by `from_raw_unchecked`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use icu_locale::subtags::Region;
+    ///
+    /// let region = Region::from_bytes(b"us")
+    ///     .expect("Parsing failed.");
+    ///
+    /// let raw = region.into_raw();
+    /// let region = unsafe { Region::from_raw_unchecked(raw) };
+    /// assert_eq!(region, "US");
+    /// ```
+    pub fn into_raw(self) -> u32 {
+        self.0.into()
+    }
+
+    /// Constructor which takes a raw value returned by
+    /// `into_raw`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use icu_locale::subtags::Region;
+    ///
+    /// let region = Region::from_bytes(b"us")
+    ///     .expect("Parsing failed.");
+    ///
+    /// let raw = region.into_raw();
+    /// let region = unsafe { Region::from_raw_unchecked(raw) };
+    /// assert_eq!(region, "US");
+    /// ```
+    /// # Safety
+    ///
+    /// This function accepts any u32 that is exected to be a valid
+    /// `TinyStr8` and a valid `Region` subtag.
+    pub const unsafe fn from_raw_unchecked(v: u32) -> Self {
+        Self(TinyStr4::new_unchecked(v))
+    }
+
     /// A helper function for displaying
     /// a `Region` subtag as a `&str`.
     ///

components/locale/src/subtags/script.rs

@@ -48,6 +48,48 @@ impl Script {
         Ok(Self(s.to_ascii_titlecase()))
     }
 
+    /// Deconstructs the `Script` into raw format to be consumed
+    /// by `from_raw_unchecked`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use icu_locale::subtags::Script;
+    ///
+    /// let script = Script::from_bytes(b"Latn")
+    ///     .expect("Parsing failed.");
+    ///
+    /// let raw = script.into_raw();
+    /// let script = unsafe { Script::from_raw_unchecked(raw) };
+    /// assert_eq!(script, "Latn");
+    /// ```
+    pub fn into_raw(self) -> u32 {
+        self.0.into()
+    }
+
+    /// Constructor which takes a raw value returned by
+    /// `into_raw`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use icu_locale::subtags::Script;
+    ///
+    /// let script = Script::from_bytes(b"Latn")
+    ///     .expect("Parsing failed.");
+    ///
+    /// let raw = script.into_raw();
+    /// let script = unsafe { Script::from_raw_unchecked(raw) };
+    /// assert_eq!(script, "Latn");
+    /// ```
+    /// # Safety
+    ///
+    /// This function accepts any u32 that is exected to be a valid
+    /// `TinyStr8` and a valid `Script` subtag.
+    pub const unsafe fn from_raw_unchecked(v: u32) -> Self {
+        Self(TinyStr4::new_unchecked(v))
+    }
+
     /// A helper function for displaying
     /// a `Script` subtag as a `&str`.
     ///